CN101116275A

CN101116275A - Quality based data scheduling

Info

Publication number: CN101116275A
Application number: CNA2005800479937A
Authority: CN
Inventors: N·约翰森; P·拉森; M·迈耶
Original assignee: Telefonaktiebolaget LM Ericsson AB
Current assignee: Telefonaktiebolaget LM Ericsson AB
Priority date: 2005-02-10
Filing date: 2005-02-10
Publication date: 2008-01-30
Anticipated expiration: 2025-02-10
Also published as: US20080144552A1; AR052667A1; WO2006085801A1; EP1851889A1; EP1851889B1; CN101116275B; US8351334B2

Abstract

The present invention relates to quality-based scheduling of data in wireless networks (1). In this scheduling, quality information (30) representing the degree of decodability of previously transmitted but not correctly received and not successfully decoded data packets (10) is estimated in receiving communications nodes (200). This quality information (30) is reported back to the node (100) that transmitted the packet (10). The quality information (30) will be used by the transmitting node (100) when scheduling subsequent data transmissions. In this scheduling process, at least one of selection of: I) receiving node(s) (200), to which a second data packet (20) is to be forwarded; ii) a type of the data in the second packet (20); and/or iii) a data flow, to which the second packet (20) belongs, is performed based on the quality information (30).

Description

Quality-based data scheduling

Technical Field

The present invention relates generally to forwarding data packets in a communication network, and in particular to quality-based scheduling of data packets in such networks.

Background

Protocols for efficiently sharing a wireless medium among multiple users are generally referred to as multiple access protocols, channel access schemes, or medium access schemes. Multiple access protocols can be divided into two main categories: conflict-free protocols and contention-based protocols.

A collision-free protocol ensures that a transmission is successful each time it is made, i.e. not interfered by other transmissions. Collision-free transmission can be achieved by statically or dynamically allocating channels to users. This is often referred to as fixed and dynamic scheduling, respectively. The benefit of accurate coordination between stations is believed to provide high efficiency, but this is achieved at the cost of complexity and sometimes the exchange of a large amount of control traffic.

The principle difference between contention-based protocols and collision-free protocols is that transmissions tend to be less successful. Therefore, the protocol should specify a procedure to resolve these issues when data is either conflicted or unsuccessfully transmitted, so that all messages are eventually successfully transmitted.

A typical example of such a collision resolution procedure is a request retransmission in the form of an automatic repeat request (ARQ). In an ARQ system, a receiving node detects whether a received data block or packet contains errors, i.e. can be successfully decoded. When an error is detected, a retransmission request in the form of a Negative Acknowledgement (NACK) identifier is forwarded to the transmitting node. On the other hand, an Acknowledgement (ACK) identifier is returned in response to successful, i.e., error-free, packet reception and decoding. Thus, error-free conditions can be achieved, but at the expense of long transmission delay times due to retransmissions, sometimes many.

An improvement of the (simple) ARQ scheme is a repeated data packet transmission in the form of Hybrid ARQ (HARQ). The HARQ scheme may employ information from previous erroneous or unsuccessful transmissions in order to improve the probability of decoding subsequent data transmissions (retransmissions).

As is known in the art, a NACK is typically returned to the transmitting side if the receiving side cannot correctly decode the original data packet. The soft values of the original data packet may be stored in memory on the receiving side, if possible. In response to the NACK, the transmitting side will retransmit the data packet (in Chase Combining (CC) HARQ scheme) or transmit incremental redundancy data (in Incremental Redundancy (IR) HARQ scheme). The receiving side can then use the soft values of the original data block and combine them with the soft values of the currently received data packet in order to increase the probability of successful decoding.

In an extension of HARQ, called reliability-based HARQ (RB-HARQ) [1,2], a simple NACK is exchanged by information identifying the weakest (least successfully decoded) bit in the original data packet. Therefore, the retransmission data packet only needs to contain these weakest bits, resulting in a smaller retransmission size. In patent application [3], an adaptive HARQ scheme is employed in which the size of redundancy transmitted in response to notification of failed decoding is determined based on a quality estimate of an unsuccessfully received data block.

However, while the RB-HARQ scheme [1,2] and the adaptive HARQ scheme [3] reduce the amount of retransmitted data, they cannot increase the throughput of a wireless communication network by providing scheduling of data packets with opportunities that enable signal transmission to be more successful than at other times and conditions. These opportunities often arise due to changes or fluctuations in the network over time. Opportunistic routing 4-6 mitigates rapidly changing link quality in the network by exploiting the opportunistic window part provided by these fluctuations. While the existing opportunistic routing schemes [4-6] work well in most cases, they are not particularly suited for HARQ based data communication. Therefore, there is a general need to provide an efficient opportunistic data packet routing and scheduling scheme suitable for HARQ based communication.

Disclosure of Invention

The present invention overcomes these and other deficiencies of the prior art solutions.

It is a general object of the present invention to provide an efficient mechanism for forwarding data packets in a wireless network.

It is another object of the invention to provide quality based scheduling of data packets in a wireless network.

It is yet another object of the present invention to provide packet scheduling that utilizes information transmitted by previous packets in the scheduling of subsequent transmissions.

It is a particular object of the invention to provide quality-based scheduling suitable for multihop networks.

It is another specific object of the invention to provide joint HARQ and (opportunistic) scheduling in a multihop network.

These and other objects are met by the present invention as defined in the appended patent claims.

Briefly, the present invention comprises providing quality information associated with unsuccessfully decoded data packets and indicative of a degree of decodability of the data packets. Such quality information is used by the transmitting node when performing scheduling of data packet forwarding.

Thus, in a main aspect of the invention, at least one communication node has received at least one data packet in one or more transmissions from one or more transmitting nodes. The communication node processes the data packet in order to attempt to decode the packet. If not, the node informs the transmitting node that the decoding failed. According to the invention, such decoding failure notification comprises quality information, called Data Quality Indicator (DQI), which indicates the degree of decodability of at least one unsuccessfully decoded data packet at the communication node. The transmitting node then performs scheduling of subsequent data packets based on this DQI(s). By utilizing this DQI (previously transmitted information) in the scheduling, the performance of the network in terms of throughput, latency, capacity and energy/power consumption can be improved.

DQI-based scheduling can be implemented according to one of many possible embodiments of the invention. In a first preferred embodiment, the transmitting node selects one or more, i.e. at least two, suitable receiving nodes from the set of multiple candidate receiving nodes based on DQI. In a preferred implementation of this embodiment, the transmitting node selects the receiving node among those candidate receiving nodes that have returned DQI. In this embodiment, the quality of the incompletely decoded information may be used for selection of the best receiving node. Thus, the transmitting node may select one or more communication nodes having a higher degree of decodability of the first packet, i.e. only a few weak or non-encodable information bits, as receiving nodes. In such a case, only a smaller amount of information has to be forwarded in the second transmission than if a communication node with a lower degree of decodability was selected.

In an alternative embodiment, the transmitting node selects, based on the DQI, the type of data in the data packet to be transmitted to the communication node, wherein the data is related to data contained in at least one of the unsuccessfully decoded data packets on which the DQI was generated. In a typical implementation, the transmitting node may select, based on DQI, whether the current data packet is: i) Is a copy of the previous packet (data retransmission); ii) contains some information bits as the previous packet and optionally some incremental bits; iii) Incremental redundancy data is (only) included. Such different alternatives may be advantageous in different situations as determined by the specific DQI values. In general, alternative i) is the best choice for low DQI values (low degree of decodability), while alternative iii) may be a good choice for high DQI values (representing a higher degree of decodability of the previous packet).

In a further embodiment, the transmitting node may select a flow among a plurality of flows represented in the transmitting node based on the DQI. Data packets are provided according to the selected flow and forwarded to at least one communication node. According to this embodiment, the transmitting node does not have to forward the data packet at the head of line of its transmit queue first. For example, the transmitting node may schedule and transmit shorter data packets first. The transmitting node may then use the DQI measure to estimate the required length of the corresponding required retransmission and to select the most appropriate data flow and packet.

Any of these embodiments may be combined to form a joint DQI-based selection. Additionally, transmission and/or link parameters used in forwarding the data packet may be determined based at least in part on the DQI. The transmitting node forwards the selected data (type) to the selected receiving node based on the selected transmit and/or link parameters.

The quality based scheduling according to the present invention may also be based on other quality information than DQI. Such additional information includes link quality data (e.g., average and/or instantaneous channel quality), the (average) routing costs associated with the transmitting node and/or the receiving node, quality of service (QoS) data associated with the different data flows represented in the transmit queue, queue status, remaining battery power, etc.

It is desirable to select the receiving node, flow and/or data type, and optional transmit/link parameters, which are optimal in some sense. In order to be able to discuss optimality in a well-defined manner, an objective function based on DQI data and optionally cost progress, link quality and QoS data is preferably introduced and optimized for the receiving node, flow and/or data type.

In a preferred embodiment of the invention, DQI data is generated not only by addressed communication nodes, but preferably also by communication nodes listening for data transmissions. In this case, the transmitting node will obtain DQI data from both the addressed (target) node and the listening (non-target) node, thus generally providing a larger selection profile that is particularly desirable in a multi-hop network.

In addition, a communication node listening for DQI reports between another communication node and the transmitting node preferably stores this DQI data for later use. Thus, the communication node may subsequently obtain responsibility for forwarding the data. If it then listens for and stores DQI generated on other nodes based on the decodability of that data, it may employ this DQI(s) in performing data scheduling, for example by selecting the most appropriate receiving node and/or the most appropriate data type or format.

According to the present invention, different quality metrics may be used as DQI to indicate the degree of decodability of unsuccessfully decoded data packets at the communication node. In general, these DQI measures should be viewed as a priori knowledge of the amount of information of a data packet that has been (previously) successfully received and processed by the communication node, and thus can represent the remaining redundancy or effort required by the transmitting node in order to successfully communicate and decode the information at the communication node.

The present invention provides the following advantages:

-increased network performance and capacity;

-increased throughput and/or reduced delay and latency;

-reduced transmission power consumption for the same performance indicators as the other schemes;

can be used to improve efficiency in multi-hop forwarding; and

can be used as a complement to existing diversity forwarding schemes.

Other advantages offered by the present invention will be appreciated upon reading the following description of embodiments of the invention.

Drawings

A more complete understanding of the present invention, as well as further objects and advantages thereof, will be obtained by reference to the following description taken in conjunction with the accompanying drawings, in which:

FIG. 1 is a flow chart illustrating a method of forwarding data packets according to the present invention;

fig. 2 is a flow chart illustrating one embodiment of the DQI receiving step of the packet forwarding method of fig. 1;

FIG. 3 is a flow chart illustrating an embodiment of the scheduling step of FIG. 1;

FIGS. 4A-4D illustrate one embodiment of data communication in a communication network in accordance with the present invention;

FIGS. 5A-5D illustrate another embodiment of data communication in a communication network in accordance with the present invention;

FIGS. 6A-6F illustrate yet another embodiment of data communication in a communication network in accordance with the present invention;

FIG. 7 is a flow chart illustrating one embodiment of the scheduling step of FIG. 1;

fig. 8 is a diagram illustrating data communications according to the present invention adapted for multi-user diversity forwarding;

fig. 9 is a diagram illustrating data communications according to the present invention adapted for selection diversity forwarding;

FIG. 10 is a schematic block diagram illustrating one embodiment of a transmitting node in accordance with the present invention;

FIG. 11 is a schematic block diagram illustrating a transmit queue of the transmitting node of FIG. 10;

fig. 12 is a schematic block diagram illustrating another embodiment of a transmitting node in accordance with the present invention;

FIG. 13 is a schematic block diagram illustrating one embodiment of a communication node in accordance with the present invention; and

fig. 14 is a schematic block diagram illustrating another embodiment of a communication node in accordance with the present invention.

Detailed Description

The same reference numbers will be used throughout the drawings to refer to corresponding or similar elements.

The present invention relates to quality-based scheduling and/or routing of data packets in wireless communication networks and systems. The basic concept of the invention is to utilize information of previous data transmissions in the scheduling of subsequent transmissions. The invention is particularly suited for providing joint HARQ and (opportunistic) scheduling in such networks.

In a main aspect of the invention, at least one communication node has received at least one data packet in one or more transmissions from one or more transmitting nodes. The communication node attempts to decode the data packet. If not, the node notifies the transmitting node of the decoding failure. In the prior art, this notification typically takes the form of a simple NACK. However, according to the invention, the communication node generates quality information associated with at least one unsuccessfully decoded data packet. Such quality information, referred to herein as a Data Quality Indicator (DQI), represents a degree of decodability of at least one unsuccessfully decoded data packet at the communication node. The transmitting node then performs scheduling of subsequent data packets based on this DQI(s).

The DQI measure can be seen as a priori knowledge of the amount of information of a data packet (previously) that has been successfully received and processed by a decoder of the communication node. DQI thus represents the remaining redundancy or effort required by the transmitting node for successful delivery of information and decoding at the communicating node.

In this respect, conventional NACKs may be considered to have been classified as a fine-grading criterion, in which DQI may be considered as a quality confirmation between conventional ACKs and NACKs. Thus, DQI can be schematically represented by a value between 0 and 1, where a value of 1 indicates ACK and successful reception and decoding of a data packet, and 0 then indicates that no information bits of a data packet can be successfully processed, i.e. decoded, at the receiving node (with the required minimum probability).

In this context, the decodability of information bits of a received data packet generally requires that the probability estimate for the bits exceed a minimum required probability threshold. For example, when decoding information bits, associated decoding-related probability values are estimated and compared to thresholds. If the threshold is exceeded, the bit is considered successfully processed (decoded) and, correspondingly, if the estimate is below the threshold, unsuccessfully decoded. This discussion may be extended directly to an entire data packet containing multiple bits of information. Thus, a data packet is considered successfully decoded if all or at least a minimal portion thereof containing information bits are successfully decoded.

Different DQI measures can be used according to the invention depending on e.g. the relevant communication network and the employed coding scheme. One example of a measure that can be used as a measure of the decodability of the DQI according to the invention is to estimate the remaining amount of information needed to decode the whole packet, i.e. the difference between the available information after decoding and the amount of information contained in the data packet processed by the decoder. Furthermore, for a flat AWGN (additive white gaussian noise) channel, the signal-to-noise ratio (SNR) is a valid indicator of the reception quality of the data packet. If trellis coding is used, the accumulated path metric represents the reception quality and can be used as DQI metric according to the invention. Accordingly, in turbo coding, A Posteriori Probability (APP) values (or a priori, intrinsic, and extrinsic information) may be used as an indication of how well bits are decoded. The a priori information is probabilistic knowledge of the bits (that can be fed into the decoder for improved decoding performance). Intrinsic information is the soft bit information sent to the decoder and extrinsic information is new information generated by the decoder. It has been shown [7] that the Probability Distribution Function (PDF) of extrinsic information is suitably represented by a gaussian shaped curve, which generally improves the mean value (essentially moving the entire gaussian bell shape) for each iteration. However, the total information is composed of not only extrinsic information but also the eigen (soft bits input from the demodulator) of the bit value and a priori information. The mean (and to some extent also the variance) of the PDF of the total information is an indication of the reliability of the decoded bits. Thus, the DQI measure may be, for example, the average total information (and optionally the variance). Another alternative is that the receiver indicates the remaining effort, i.e. the remaining required information, instead of the amount of received information. The remaining required information is then the difference between the total information received and the required amount of information that allows full decodability with a certain probability. Alternatively, the receiver may map the remaining required information knowledge to the appropriate HARQ transmission format, which will then be used as DQI metric.

The parameters listed above should be considered as typical, but non-limiting examples of suitable DQI measures that can be used according to the invention. Other possible metrics may include the percentage of decodable information bits in the packet, i.e., primarily referring to values between 0 (0.0) and 100% (1.0). Accordingly, an indication of weak bits that are unsuccessfully decoded (e.g., by having an associated probability estimate less than a minimum threshold) is considered a DQI metric in accordance with the invention.

In general, any quality measure representing the degree of decodability of unsuccessfully decoded data packets at the communication node may be used as DQI measure.

Fig. 1 is a flow chart illustrating a method of forwarding data packets according to the present invention employing quality-based scheduling.

The method starts at step S1, where the transmitting node receives DQI (quality information indicating the degree of decodability of unsuccessfully decoded data packets). A transmitting node may receive one or more DQIs from one or more communication nodes. These DQIs can be associated with the same data packet or different data packets. This data packet(s) may have been previously transmitted by the transmitting node. Alternatively, at least a portion of the DQI may instead be associated with data packets already transmitted by other nodes, as discussed in more detail below.

In a next step S2, the transmitting node performs scheduling of data packet forwarding or transmission based on the DQI. The DQI may be the DQI received in step S1, a part thereof and/or a DQI previously received and stored in the transmitting node. In either case, DQI-based scheduling preferably takes advantage of the opportunities provided by system variations and fluctuations. By further exploiting previously transmitted information via the use of DQI in scheduling, the performance of the system in terms of throughput, latency, capacity and energy/power consumption can be improved.

DQI-based scheduling can be implemented according to one of a number of possible embodiments of the invention, discussed in more detail below. In short, the transmitting node may select one or more, i.e. at least two, suitable receiving nodes based on the DQI. In a preferred embodiment of the invention, the receiving node is preferably selected among the communication nodes that have returned DQI, based on DQI. However, additionally or alternatively, at least one communication node that does not return a DQI value can be selected as a receiving node, as long as the selection process is based at least in part on the at least one DQI value. Alternatively, the transmitting node selects, based on the DQI, a type of data in the data packet to be transmitted to the communication node, wherein the data relates to data contained in at least one of the unsuccessfully decoded data packets on which the DQI was generated. Furthermore, the transmitting node may select a flow among the plurality of flows represented in the transmitting node based on the DQI. Data packets are provided according to the selected flow and forwarded to at least one communication node.

Any of these embodiments may be combined to form a joint DQI-based selection. Additionally, transmission and/or link parameters used in forwarding the data packet may be determined based at least in part on the DQI.

In a next step S3, the transmitting node forwards the data packet according to the performed DQI-based scheduling. Note that the forwarded data packet preferably carries data related to at least a part of the data contained in at least one of the unsuccessfully decoded data packets, from which the DQI has been determined. Thus, the forwarded data packet may be a (identical) copy of the unsuccessfully decoded packet previously transmitted. Alternatively, the data packet may carry partially identical bits and partially incremented bits relative to the previous transmission. In yet another example, the data packet carries only incremental redundancy bits associated with the previous packet.

The method then ends.

The forwarding method of fig. 1 can be used as an extension of conventional automatic repeat request techniques, such as HARQ schemes, wherein the use of received DQI data to determine at least one of the following is inventively added: i) To whom the subsequent data packet is sent, ii) the type of subsequent data packet, and iii) which flow and data packet is forwarded first.

Fig. 2 is a flow chart illustrating an embodiment of the receiving step S1 of fig. 1. In a first step S10, the transmitting node forwards a first data packet destined for at least one communication node. This first data packet may have been generated in the transmitting node from a higher layer (application) or received from another node in the system. The data packet may be: unicast, i.e., addressed and forwarded to a single node; multicast, i.e. addressed and forwarded to a plurality of nodes; or broadcast, and typically forwarded to one or more non-addressed nodes. In the latter two cases this can be done, for example, by direct forwarding without detecting the channel or by employing CSMA (carrier sense multiple access) or CSMA/CA (CSMA with collision avoidance).

The node processes the first data packet in order to attempt to decode it. The communication node preferably generates and returns an ACK if the decoding is determined to be successful, for example by a test via a CRC (cyclic redundancy check) check. If it is determined to be unsuccessful, e.g. by a CRC check, the node preferably generates and returns a DQI reflecting the decodability of the first packet on that node at a next step S11.

Corresponding to the forwarding of the first packet, the communication node may return the DQI by unicast, multicast or broadcast. In particular, in the latter two cases, the DQI will or may be received by other nodes than the transmitting node. It may be advantageous if these other nodes store this DQI information, since they can then use the DQI when (later) responsible for further forwarding the first data packet.

DQI can be returned directly after its estimation or determination at step S11. Alternatively, DQI is returned after a predetermined period of time. In yet another embodiment, the DQI is not returned to the transmitting node until a report request is received from the transmitting node. Alternatively or additionally, particularly for multicast or broadcast implementations, the DQI may be continuously or periodically transmitted from the communication node.

The DQI metric can be estimated or determined by the relevant communication node. Alternatively, the communication node provides (quality) information and transmits it to the transmitting node. This transmitting node may then calculate or at least estimate DQI from the received quality information. The specific solution to be employed depends on where the processing complexity is to be increased and the generality that the DQI will have.

It is further noted that in a preferred embodiment of the invention, not only the addressed communication node generates and transmits DQI. If the communication node overhears the first data packet transmission, DQI may be generated (if decoding of the overheard packet is unsuccessful) and returned to the transmitting node.

This means that a given transmitting node may have access to DQI associated with data packets it itself transmits (and that are present in the transmit queue) and DQI associated with data packets transmitted by other nodes, e.g. stored in the transmit queue or other memory area. These other DQIs can be used if the transmitting node is to subsequently receive and be responsible for further forwarding of data. These previously received DQIs can then be used in scheduling.

The method then proceeds to step S2 of fig. 1.

Fig. 3 is a flow chart illustrating different embodiments of DQI-based scheduling of step S2 in fig. 1.

In a first embodiment represented by step S20, the transmitting node has obtained DQI from at least one communication node in a set of multiple candidate receiving nodes. It then selects an appropriate receiving node from the set of candidate nodes based on the DQI for forwarding the (second) data packet to it. Thus, in this selection step S20, the transmitting node may select a receiving node from the set that has returned a DQI value or that has not returned a DQI, based on the reported DQI. The latter case may for example be due to the receiving node being added to the candidate set after transmission of the previous (first) data packet (e.g. because the path loss has changed). It is also possible to select in step S20 and on the basis of DQI a number of receiving nodes that have returned or have not returned DQI values. In a further extension, the transmitting node selects, based on DQI, at least one receiving node that has returned DQI and at least one receiving node that has not returned DQI, step S20.

However, in a preferred implementation of this first embodiment, the selected receiving node has reported DQI data to the transmitting node. In a further preferred implementation, the transmitting node receives DQI data from a plurality of communication nodes of the candidate set. In step S20, one or more receiving nodes, and optionally also DQI-unreported nodes, are then selected from those communication nodes that have reported a set of DQIs based on DQI.

For example, assume that a transmitting node is disposed in a multi-hop communication network and will forward a data packet to a destination node via multi-hop transmission. This data packet has been sent to the first communication node arranged in the direction of the final destination node. A second communication node, disposed in proximity to the first node, listens for packet transmissions. It is assumed that both the first and second nodes are unsuccessful in fully decoding the data packet, thus returning a DQI value to the transmitting node. In this illustrative example, the DQI from the addressed first node is 0.4 (where a value of 1 indicates a successfully decoded packet and 0 indicates a completely unsuccessful decoding, i.e. no decodable information bits at all). The corresponding DQI value from the listening second node is 0.8. The transmitting node may then select one of the two nodes to retransmit the data packet or to transmit the relevant redundant data thereto. Since the second node has the ability to successfully process a larger portion of the data packet than the first node, less data typically needs to be transmitted and less energy is spent if subsequent data packets are forwarded to the second node instead of the first node.

This should be compared to the prior art, where the subsequent data packet is only transmitted to the first node, despite the higher degree of decodability in the second node.

The invention is therefore well suited for scheduling and routing in multihop networks. In general, the multi-hop approach provides several advantages, such as lower power consumption and higher information throughput than the direct single-hop approach. In a multihop network, nodes that are unreachable to each other may benefit from intermediately located nodes that may forward messages and packets from a source to a destination using routes determined at least in part based on DQI metrics.

In a second embodiment, shown in step S21, the transmitting node determines the type of data to be transmitted in the current data packet based on the DQI associated with the previous packet containing at least part of the data related to the data in the current packet. In a typical implementation, the transmitting node may select whether the current data packet is: i) Is a copy of the previous packet (data retransmission); ii) contains some information bits in common with the previous packet and optionally some incremental bits; iii) (ii) contains incremental redundancy data; iv) some other alternative. Such different alternatives may be advantageous in different situations as determined by the specific DQI value.

For example, assume that the communication node receives a first data packet from the transmitting node. The radio conditions during data forwarding are very poor and thus the communication node may only successfully process (with sufficient probability) a small fraction of the first packet, represented by a DQI value of 0.1. A corresponding situation is assumed but with better radio conditions, so that a large part of the packet is decodable by the communication node resulting in a DQI value of e.g. 0.85. Comparing these two examples, it may be most efficient to transmit redundant data only in the second data packet in case of having an almost fully decodable packet (DQI = 0.85); whereas in case of poor decodability (DQI = 0.1), the second data packet will be a duplicate of the first packet. Furthermore, aspects such as average channel quality (e.g. in the form of signal-to-noise-and-interference ratio (SINR)) or average channel gain may be taken into account together with the DQI value for optimal selection. In addition to the average channel quality or gain, their instantaneous values may also be considered or considered in conjunction with the corresponding average value, discussed in more detail below.

The present discussion may be extended such that the transmitting node is configured to provide a second data packet when 0 ≦ DQI < α ₁ When it is a copy of the first packet, when α ₁ ≤DQI＜α ₂ Including a portion of the information bits of the first packet and optionally a portion of the incremental bits, anWhen alpha is ₂ When the DQI is less than or equal to 1, only incremental redundancy is carried, wherein, alpha is more than 0 ₁ ＜α ₂ < 1,0 ≦ DQI < 1, and high DQI values indicate a high degree of decodability.

In a third embodiment, shown at step S22, the transmitting node maintains a list of destinations/flows currently represented in the node. DQI-based scheduling provides for selection among these different destinations/flows. In practice, this enables the selection of the receiving communication node in several general forwarding directions.

For example, assume that the first data in the transmit queue (i.e., the head-of-line packet of the queue) is associated with DQI =0.2 received from a first communication node disposed in a first direction relative to the transmitting node. This first data is intended for a first target node in a first direction. The transmit queue further comprises second data associated with DQI =0.9 received from the second communication node and belonging to the second flow. This second node may be arranged in a second general direction between the transmitting node and a second target node to which the second data is to be sent.

Applying the prior art, the transmitting node first transmits a data packet associated with the first data, since this data is first in the queue. However, scheduling short data packets first is a useful scheduling strategy to reduce the average queuing latency. According to the present invention, it may be most advantageous to transmit the data packets associated with the second flow first, since such packets may typically be smaller due to the higher associated DQI value. Scheduling strategies other than selecting short packets may also be applied, such as optimizing other metrics that represent progress to the destination and the effective amount of information to deliver.

Therefore, DQI-based scheduling according to the present invention can be performed according to the above first embodiment (step S20), second embodiment (step S21) or third embodiment (step S22). In an additional preferred embodiment, a joint DQI-based selection is performed, combining the procedures of at least two of the steps S20 to S22. In other words, the transmitting node may choose to which node to forward the packet and what data type the packet will contain, based on the DQI (steps S20 and S21). Alternatively, the transmitting node selects a data flow and a receiving node in the flow direction based on the DQI (steps S20 and S22). In addition, steps S21 and S22 may be combined such that the transmitting node selects, based on DQI, a data flow and a packet and the type of data this packet should contain. Yet another embodiment of the scheduling step comprises a selection of a data flow, a selection of a receiving node in the direction of the flow and a selection of a type of data forwarded to the selected receiving node (steps S20, S21 and S22), wherein all these selections are performed at least partly on the basis of DQI. The other (preferred) input in the DQI-based selection is the average or instantaneous channel quality (or pure path gain). In addition, qoS parameters may also be considered in the scheduling process, which is discussed further herein.

Any of the above embodiments (single or joint DQI-based selection) may be combined with selecting transmission and/or link parameters for data packet forwarding in step S23 based on DQI. The transmission parameters include, for example, transmit power and antenna weights used in data forwarding. Other parameters, such as coding and rate or link parameters, including link mode parameters, e.g. data signal constellation and forward error correction coding or frequency channel parameters, may be adjusted and selected based on the obtained DQI values.

For example, assume that two communication nodes return DQI values of 0.2 and 0.9, respectively, in response to unsuccessful reception of a first data packet from the transmitting node. The transmitting node may then adjust the transmission parameters based on the received DQI value in order to increase the probability of successful communication (retransmission) between the transmitting node and at least one of said nodes. Therefore, lower transmit power may typically be employed when forwarding the second data packet (e.g. retransmitting the first packet) to a communication node having a DQI value of 0.9 compared to forwarding the packet to a further node.

The transmit parameters may also be adjusted based on DQI to maximize overall efficiency, for example, by reducing unnecessarily high transmit power levels to reduce energy consumption and other unwanted side effects. The transmission parameters may be set such that there is a high probability that the transmission reaches the closest communication node and also a probability that the transmission reaches one or more nodes further in the propagation direction. If one of the further nodes is reachable, there may be fewer hops in a multi-hop journey of data through the network, which will save overall energy consumption.

The method then proceeds to step S3 of fig. 1, in which the (selected) data packet comprising data of the (selected) type is transmitted to the (selected) receiving node using the (selected) transmission and/or link parameters.

The invention will now be further illustrated by way of example, wherein DQI-based scheduling in the form of receiving node selection is mainly described. However, as noted above, other embodiments or combinations of the scheduled embodiments may be employed.

Fig. 4A-4D illustrate data communication in a multihop communication network 1 according to the present invention. In fig. 4A, the transmitting node 100 will transmit a Data Packet (DP) intended for a destination node by using an intermediate communication node. The transmitting node 100 typically first selects a set of suitable candidate relay or receiving nodes 200-1 to 200-3 arranged in the propagation direction towards the target node, i.e. these selected nodes 200-1 to 200-3 constitute the set of candidate nodes. The data packet is then forwarded to these candidate nodes 200-1 to 200-3 using, for example, broadcast or multicast techniques. An advantage of employing a set of multiple candidate nodes 200-1 through 200-3 is that the probability of at least one of the candidate nodes 200-1 through 200-3 successfully receiving a data packet is generally increased as compared to using a single candidate node.

The relay candidate nodes 200-1 to 200-3 attempt to decode the data packet, for example, by testing the data via CRC checking. However, in this example, none of the candidate nodes 200-1 through 200-3 successfully decode the data packet. Preferably, each of the nodes 200-1 to 200-3 generates DQI, or at least information from which DQI can be estimated, wherein DQI represents the degree of decodability of a data packet on that particular node 200-1 to 200-3. In fig. 4B, DQI is reported to the transmitting node 100, here represented by an addressed transmission.

In fig. 4C, the transmitting node 100 performs scheduling based on the obtained DQI. Such scheduling may be achieved by selecting the candidate node that returns the largest DQI, i.e. has the highest degree of decodability of the data packet. However, in other embodiments, other parameters, including (instantaneous and/or average) link quality, routing costs, quality of service (QoS) data, queue status, remaining battery power, etc., may be used for scheduling, possibly resulting in selecting one of the non-largest DQI-associated nodes being more efficient, as discussed in more detail below. The transmitting node may provide a second data packet (DP') based on the DQI. This second data packet preferably comprises data related (identical or providing additional redundancy) to the data in the first packet (DP). This second packet may be sent to the relay node 200-2 selected based on DQI, possibly with transmission parameters selected based on DQI.

The relay node 200-2 may then use the first Data Packet (DP), which is a stored soft value of the information bits in the first Data Packet (DP), and the second packet (DP') in the decoding process, greatly increasing the probability of successful decoding. In this example, the decoding may be performed correctly and the relay node 200-2 returns an ACK. The transmitting node 100 preferably deletes the relevant data, i.e. the first and second packets, from its transmit queue in response to the reception of the ACK. The other candidate nodes 200-1, 200-3 may also delete data packets from their storage if they listen for this ACK transmission or if the ACK is multicast/broadcast. In an alternative embodiment, the data packet is deleted upon the elapse of a predetermined period of time or upon receipt of a drop command.

Fig. 5A to 5D illustrate another example of DQI-based scheduling and data forwarding in a multihop network 1. In fig. 5A, the transmission node 100 transmits a Data Packet (DP) to the relay node 200-1. The second relay node 200-2 listens for this initial packet transmission. Two relay nodes 200-1, 200-2 attempt to decode the data packet but are unsuccessful. They generate DQI reflecting the respective degree of decodability of their resulting packets and return it to the transmitting node 100 in fig. 5B.

The transmitting node selects a relay node among the two candidate nodes 200-1, 200-2 based on the two received DQI values, which will receive a second data packet (DP'), which may be a duplicate of the first packet or contain redundant data. In this particular example, listening node 200-2 is the selected relay node and thus the transmitting node 100 forwards the second packet to it in fig. 5C. In this second data transmission, the relay nodes 200-1, 200-3 listen for transmissions, corresponding to fig. 5A.

All of these relay nodes attempt to decode the data, the relay nodes 200-1, 200-2 proceed by employing the first (DP) and second (DP ') packets and the relay node 200-3 employs only the second packet (DP'). The relay node 200-1 successfully decodes and generates and returns an ACK, while the other two nodes 200-2, 200-3 may process only a portion of the data, thus generating and returning DQIs in fig. 5D.

Fig. 6A-6F illustrate yet another example of a multihop network 1 employing the present invention. The transmitting node 100 has data in its transmit queue that it will forward to the destination node in a multi-hop fashion. The node 100 provides a set of suitable candidate nodes 200-1, 200-3 that may serve as intermediate relay nodes for multi-hop transmissions. Data packets comprising relevant data are compiled and transmitted, preferably multicast, in fig. 6A to the candidate nodes 200-1, 200-3. Another relay node 200-2 not included in the candidate set listens for this packet transmission.

In fig. 6B, one of the candidate nodes 200-1 successfully decodes the data and returns an ACK to the transmitting node 100. However, the remaining candidate nodes 200-3 and listening node 200-2 may not fully decode the packet and provide and return DQIs reflecting their respective degree of decodability. In this example, the transmitted DQI is listened to by the relay node 200-1, or the DQI may be broadcast or multicast. In either case, the relay node 200-1 stores this DQI information in a memory, e.g., in its transmit queue.

Upon receiving the ACK and DQI, the transmitting node 100 determines that the first relay node 200-1 (due to successfully receiving the data) will be responsible for subsequent forwarding of the data to the target node. Accordingly, the transmitting node 100 sends a forwarding instruction (FO) to the relay node 200-1 in fig. 6C. In response to this forwarding instruction, the relay node returns a forwarding ACK to the transmitting node in fig. 6D. The transmitting node 100 may then remove the data from its transmit queue and release all responsibility for the data.

In fig. 6E, the previous relay node (200-1 in fig. 6A to 6D) is considered as the second transmitting node 100-2, since it has the responsibility to forward the data towards the final destination. In contrast to the situation of fig. 6A, this second transmitting node 100-2 has a priori knowledge (DQI) that a part of the data has been previously received by the relay nodes 200-2, 200-3. Thus, DQI information may be used when selecting a candidate node among these nodes 200-2, 200-3. In addition, the DQI information may be used to determine the type of data to be transmitted to the candidate node. For example, if the DQI overheard from the relay nodes 200-2 and 200-3 (see fig. 6B) have values of 0.9 and 0.4, respectively (0 ≦ DQI < 1), the transmitting node 100-2 may choose to forward only a portion of the original data packet or redundant data related to the packet previously received by the nodes 200-2, 200-3 in fig. 6A, based on DQI. Therefore, less information needs to be sent than in the prior art case where the transmitting node 100-2 does not have a priori knowledge about the magnitude of the residual redundancy required for successful decoding. The transmitting node provides a data packet (DP') including this portion or redundancy of the data and transmits to the candidate node 200-2. Another relay node 200-3 listens to this transmission.

The two nodes 200-2, 200-3 preferably attempt to decode the data using the currently received data (DP') and the previously received Data (DP). In this example, the candidate node 200-2 may decode the data correctly while the other node 200-3 is still unsuccessful. Thus, ACK and DQI are returned to the transmitting node 100-2 in fig. 6F.

In the above example, the transmitting node has selected a (single) receiving or candidate node based on the DQI value. However, the present invention contemplates that a plurality of suitable receiving or candidate nodes may alternatively be selected based on the DQI value.

Fig. 7 is a flow chart illustrating in more detail one embodiment of the scheduling step S2 of fig. 1. In this embodiment, other data and parameters than DQI are also used for the scheduling procedure. The method continues from step S1 in fig. 1.

In a next step S30, the transmitting node transmits a channel sounding or interrogation message to a plurality of possible receiving nodes. Information of possible relays or candidate nodes may be derived from earlier derived topology information, but may also be affected by transmit queue content, DQI and QoS factors. This topology and connectivity data may be collected and stored relatively slowly. For example, such data collection may be a continuous process. The topology and connectivity data may then be used to help the transmitting node determine which nodes are appropriate candidates when the forwarding process is activated. The collection rate of topology and connectivity data is ideally: i) High enough to provide each node with a general indication of which node is a suitable candidate node for data packets traveling in a particular direction or to the final destination, while at the same time ii) low enough to avoid wasting energy and other resources saving overhead information. The collection of topology and connectivity data may be provided via conventional routing information protocols, such as different bellman-ford algorithms.

This inquiry message employs a probe to detect the current quality of the channel or link between the transmitting node and the relay node. Each receiving node estimates its channel or link quality and generates a Channel Quality Indicator (CQI). This CQI may be an estimated SNR or SINR. SINR is preferred in most applications. In opportunistic scheduling, CQI is often considered as instantaneous channel quality, but may also be a measure of the average channel quality over some predefined period of time. When available, it is generally preferred to employ CQI data that includes both instantaneous and average channel quality. In step S31, the relay node transmits a response message including the CQI. In an alternative embodiment, each candidate node determines which rate is available for reception and then responds with that rate in step S31 instead or in addition. The rate may be an exact value or an implicit code of the modulation (QPSK, 8PSK, 16 QAM.,) and some combination of forward error correction code (convolutional coding, turbo coding,. And.) and coding rate.

At step S32, the transmitting node, when employing some form of cost progress, may provide route cost information via an independent route determination protocol, such as any well-known shortest path protocol (e.g., bellman ford or Dijkstra), for example employing energy, delay or hop metrics or a custom route determination protocol. At this step S32, the transmitting node may provide a routing cost associated with one or more of itself and/or the candidate relay nodes.

As is generally known in the art, different flows may have different QoS requirements. By way of example, flows with strict delay requirements are typically prioritized higher than flows with more relaxed delay requirements. Thus, at step S33, the transmitting node may provide QoS parameters associated with the data flows or packets represented in its transmit queue.

In the scheduling step S34, the transmitting node may employ any of the above parameters CQI, routing cost, qoS in addition to DQI. In most cases, using more decision data and different types of decision data generally results in more efficient scheduling and routing. Thus, the DQI-based scheduling of the invention is advantageously implemented as a scheduling based on DQI data and at least one of CQI, route cost and QoS.

The method then proceeds to step S3 of fig. 1.

Combining diversity scheduling

The quality based scheduling proposed by the present invention can be used together with other opportunistic or multi-user diversity scheduling or communication schemes used in cellular communication systems, e.g. Wideband Code Division Multiple Access (WCDMA) systems such as WCDMA-HSDPA (WCDMA high speed downlink packet access) or CDMA systems such as CDMA-HDR (CDMA high data rate) [8,9] or multi-hop networks [4-6 ].

Referring to fig. 8, the DQI-based opportunistic scheduling of the present invention can be combined with a multi-user diversity forwarding (MDF) scheme [5 ].

The transmitting node TX1 has in its transmit queue data that it will forward to the destination node in a multi-hop fashion using intermediate relay nodes in the network. Thus, the transmitting node compiles the data packet 10 containing the relevant data and sends it to the possible candidate relay nodes RX1 to RX3, e.g. by broadcasting or multicasting the data packet 10. However, in this example, the other transmitting node TX2 also transmits data interfering with the packet forwarding. Thus, none of the candidate nodes RX1 to RX3 can successfully decode the received packet 10. Each relay node RX1 to RX3 preferably estimates a DQI 30-1 to 30-3 that reflects the degree of decodability of the packet 10 at that node RX1 to RX3. These DQIs 30-1 to 30-3 may be returned to the transmitting node TX1 immediately after evaluation as shown, or to the transmitting node TX1 after a predetermined period of time has elapsed or upon request from the transmitting node TX1.

In MDF-oriented DQI-based scheduling, the transmitting node performs scheduling using CQI metrics in addition to DQI data. To obtain these CQI measures, the transmitting node TX1 generates an inquiry message or probe 40 and transmits it to the possible relay nodes RX1 to RX3. This is preferably also done before the previous data transmission 10, but is omitted in fig. 8 in order to simplify the drawing. Each relay node RX1 to RX3 generates CQI metrics 50-1 to 50-3, e.g. by estimating SNR/SINR, and reports the estimated CQIs 50-1 to 50-3 to the respective transmitting node TX1.

In order to ensure substantially the same interference conditions in the interrogation phase and in the subsequent data phase, the transmitting nodes TX1, TX2 should preferably transmit their frames in a time-synchronized manner, and substantially the same transmit power level and/or antenna weights should be used in both phases. As shown in fig. 8, the transmitting nodes TX1, TX2 transmit their frames in such a way that the time slots are time aligned. This provides the basis for the correlation between the interrogation phase and the data phase. In addition, one or more transmission parameters, such as transmit power level and/or antenna weights, are initially determined and preferably based at least in part on previously received DQI data 30-1 to 30-3. These transmission parameters are preferably used in the challenge phase and subsequent data phase so that the CQI 50-1 to 50-3 reported in the challenge response phase remains the same (or is improved) throughout the data phase.

The transmitting node TX1 performs scheduling because the receiving node is selected among the candidate nodes RX1 to RX3, e.g. based on the DQI 30-1 to 30-3 and the CQI 50-1 to 50-3 data. Alternatively or additionally, the transmitting node TX1 may select the type of data to be transmitted in the subsequent data phase and/or from which data stream a packet is to be transmitted first, based on the DQI 30-1 to 30-3 and CQI 50-1 to 50-3 data. In the illustrative example of fig. 8, the transmitting node TX1 selects node RX2 as the receiving node, and preferably compiles a second data packet 20 to be transmitted to the receiving node RX2 based on at least the DQI 30-2 and optionally the CQI 50-2 data from that node RX2.

In this case, the receiving node RX2 may correctly decode the data by using the previously received first packet 10 and the currently received second packet 20. The node RX2 returns an ACK 60 informing the transmitting node TX1 of the successful reception, allowing the node TX1 to delete the corresponding data from its transmit queue. The listening node RX1 captures the second packet 20 but fails to decode the information correctly and preferably returns the relevant DQI value 30.

Fig. 9 illustrates an example of data signaling in which DQI-based opportunistic scheduling of the present invention is combined with Selection Diversity Forwarding (SDF) [4 ]. In this method, the transmitting node TX1 directs the transmission of a data packet 10 to a group of receivers or relay nodes RX1 to RX3 in the vicinity. Corresponding to fig. 8, the data transmission of the further transmission node TX1 interferes with the packet forwarding, resulting in an imperfect decoding in the relay nodes RX1 to RX3. According to the invention, these nodes RX1 to RX3 report incomplete decoding by estimating and returning DQI data 30-1 to 30-3. The transmitting node TX1 selects at least one of these relay nodes RX1 to RX3 as receiving node RX2 according to the reported DQIs 30-1 to 30-3. The second data packet 20 is passed to the selected node RX2. However, the second transmitting node TX2 again interferes with packet forwarding and despite access to both the first 10 and second 20 data packets, the selected node RX2 still cannot decode the data. One of the other relay nodes RX1 listens to this second packet forwarding and successfully decodes the data correctly. Thus, the selected node RX2 reports DQI metrics 30, while the listening node RX1 returns an ACK 60.

Upon receiving the DQI 30 and ACK 60, the transmitting node TX1 transmits a forwarding instruction command 70 instructing the relay node RX1 to assume responsibility for further forwarding the data. The relay node RX1 responds by returning a forward command acknowledgement 65. This process is repeated for all subsequent responsible nodes until the information reaches the final destination. Note, however, that if, for example, one or more of the nodes RX1 to RX3 can successfully process (decode) the data after receiving the first packet 10, the transmitting node TX1 typically selects one of the successfully decoded nodes and sends a forwarding instruction to that node.

By following this approach, branch diversity and capture effects can be used in the data forwarding process. In particular, branch diversity reduces the need to use interleaved data along with coding to prevent fading channels, which in turn means less delay and thus higher throughput. The capture effect refers to a phenomenon: the stronger of the two signals at or near the same frequency is demodulated while the weaker signal is suppressed and discarded as noise. In conjunction with multiple receiving or relay nodes, the capture effect provides a high degree of robustness in the event of data transmission collisions.

In the scheduling of the present invention it is desirable to select, based on DQI data, the receiving node, the type and/or flow of data and optionally transmission and/or link parameters, which are in a sense optimal. In order to be able to discuss optimality in a well-known manner, an objective function f is usually introduced. This objective function f is carefully chosen and depends on the following: i) A given input parameter, and ii) certain variables that can be carefully selected to optimize the objective function f.

In the present invention, the input parameters comprise quality information, i.e. DQI data, indicating the degree of decodability of a previously unsuccessfully decoded data packet. Other input parameters may also be used, such as routing costs associated with the node, CQI data, qoS requirements, queuing status, or remaining battery power.

The optimization variables include the receiving node, the data type and/or data flow, and optionally the transmit/link parameters. In addition, rate may be included as a variable as desired. The rate is then determined by any suitable combination of modulation, coding and spreading schemes.

The output from the objective function f comprises at least one of: i) A selected receiving node, ii) a selected data type, iii) a selected data stream or destination. The selection of the destination or flow affects which information is to be sent. In addition, optimization of the objective function may also provide a suitable combination of modulation, coding and spreading, i.e. rate selection, and a suitable set of subcarriers or frequency channels to be used (i.e. in a multi-carrier system, such as an Orthogonal Frequency Division Multiplexing (OFDM) or Orthogonal Frequency Division Multiple Access (OFDMA) system). The reduction in transmit power is another additional output due to rate selection.

In the following discussion, the invention is further described by way of example, in which the joint selection and optimization of the best receiving or relay node and the best flow or the best node, the best flow and the most appropriate rate. However, with slight modifications, the present discussion may be applicable to a single optimization of the best relay node, the best data type, the best flow, or a combination of at least two of these objectives, possibly in combination with optimizing transmission/link parameters.

In formalizing the optimization considering the relay nodes and the flows, the following notation may be used:

v represents the set of all nodes in the network or the considered part of the network.

J _i Is a set of candidate relay nodes.

Φ _i Is node v _i Set of streams in v _i ∈V。

In this illustrative discussion, the objective function f is then for the representative node v _i Is optimized using the forwarding from set J above _i And phi _i Jointly determining relay nodes by input parameters

And flowThe optimal combination of (A):

wherein:

defining the selected relay node:

defining the selected stream:

in the first case, the only available input parameters are DQI and what data packet residue in the transmit queue. The scheduling criteria may be based on the objective function f according to the following equation:

Z _ij ＝f(DQI _j ) (4)

wherein v is _i Is a transmitting node, v _i ∈V，v _i Are possible receiving (relaying) nodes, v _i E.g. V and J e.g. J _i And  _i Is a stream,  _i ∈Φ _i 。Z _ij Relatively of a Luy _j And  _i Optimization according to equations (1) to (3), i.e. jointly finding the optimal flow

And an optimal relay node

In a particular embodiment, information about the slave node v is also obtained _i To node v _j Information of average cost of (a). As previously described, shortest path protocols such as bellman-ford or Dijkstra may be used to determine the average routing cost.

C _i ^(i) Representing a stream  _i Slave node v _i Cost to destination, v _i ∈V， _i ∈Φ _i . Each flow is associated with a certain destination.

The objective function f of equation (4) can then be modified as follows:

since it is desired that the data be moved toward the final destination,

that is, it is assumed that the cost monotonically increases from the destination.

Although equation (5) is specified by flow, it can be specified by destination as well. For example, node v _i May have a link with the target node v _d Flow of  _d Corresponding associated cost C _i ^(d) . However, if node v _i Will be seen as equal from a cost point of view, C _i ^(d) Can be composed of C _i ^(d) And (4) substitution. This means that the slave node v _i Flows destined for the same destination encounter the same routing cost. Formula (5) can then be rewritten as:

an alternative is that the (instantaneous) link conditions, represented by the CQI metric, are known. The scheduling condition can be expressed as maximizing:

Z _ij ＝f(DQI _j ，CQI _ij ) (7)

or more generally (instantaneous) link conditions and average costs:

a specific example of equation (8) is a modified form of quality cost progress (Z) ^QCP ). In the inventionMiddle, stream  _i Node v of _i And node v _j Quality Cost Progress (QCP) between is defined as:

wherein:

W _i ^(i) is node v _i And stream  _i The weighting parameter of (2).

The weighting parameters may be any combination of at least fixed prioritization weights, adaptive prioritization weights, qoS related parameters (e.g., expiration time, latency, etc.), fairness criteria, and the like. It is somewhat more natural and straightforward to incorporate QoS parameters into the optimization when considering flows as optimization variables, since each flow is typically associated with a given QoS requirement.

This allows us to write the optimization of the objective function (assumed here to be maximized) according to QCP as:

it produces a combination of relay nodes and flows. Note that if Z is _ij ^QCP max is negative, no forwarding is performed.

Another special exemplary objective function is based on information cost progression (Z) ^ICP ). The following additional symbols may be used:

R _ij is at a given SINR value CQI _ij Node v in case (2) _i And node v _i May implement a set of rates. The rate is formed by a combination of modulation, coding and spreading schemes.

In the present invention, the rate r is adopted _ij Stream  _i Node v of _i And node v _j The Information Cost Progress (ICP) between is defined as:

optimization of the ICP-based objective function (assumed here to be maximized) can be expressed as:

this results in a combination of relay node, flow and selected rate. Note that if Z is _ij ^ICP max is negative, no forwarding is performed.

In the combination of the MDF scheme with the invention, the transmitting node v _i Try to find which sectionThe point is determined to be the target of the transmission. Proportionally fairness if the head of line grouping of a single stream is consideredThe degree may work as follows.

First assume node v _i And node v _j CQI between _ij Providing instantaneous link rate r _ij And the (multi-hop) cost metric is based on the inverse of the average rate of each link:

as disclosed herein, DQI _i (it is each node v) _i And packet-specific, with another assumption being that the head-of-line packet is represented by its stream ) can take many different formats. In this example, it is proposed that the DQI parameter take the form of redundancy that enables the additional pre-metering needed for correct decoding of a (previously incorrectly received) packet. If the length of the original (previous) data packet is L _j， And L is the required retransmission with redundant bits _j， ^(red) Then the effective instantaneous rate r _ij For different possible receiving nodes v _j Will be different.

The scheduling conditions may then select a node for the head-of-line packet belonging to flow :

for each stream , a relay node and rate are selected:

this metric minimizes the expected transmission delay, which can alternatively be interpreted as maximizing the capacity of the network, since the least possible time resources are used to transmit packets. For example, if low average delay is important, the QoS aspect may be included in equation (14), e.g., by determining the priority order of flows characterized by short packets or short redundant packets.

Tentative relay node assuming selection of each flow

Sum rateThe task is to select the best stream. This can be achieved by selecting the flow that provides the largest relative rate increase relative to the corresponding average rate, i.e. proportional fair scheduling, according to the following equation:

in another alternative embodiment, the ratio between the instantaneous channel quality and the average channel quality is used as a criterion instead of equation (16). This provides fairness over the SNR range (e.g., considering the logarithmic relation of shannon capacity versus SNR), as opposed to rates that are non-linear with respect to SNR.

Summarizing equations (14) to (16), the selected relay nodes, rates and flows are:

in the previous optimization example, the DQI parameter assumes a redundancy (L) of the additional pre-measure needed to be able to correctly decode the packet _j， ^(red) ) In the form of (1). Generally, the transmit power may affect the redundancy of this expected amount. For example, suppose L _j， ^(red) Redundancy bit prediction for transmission at power level P ^j，(1) The transmitted (retransmitted) packet is successfully decoded. Accordingly, L may be required _j， ^(red2) Redundant bits to power level P _j， ⁽²⁾ The corresponding information transmitted (retransmitted) is successfully decoded. If it is used

Then in general terms the number of bits to be processed,

in other words, lower transmit power may be compensated for by transmitting more redundant data. In most practical implementations, there is a trade-off between the power level to be used and the number of redundant bits that should be transmitted. Therefore, optimization of the objective function may also focus on this power-bit size problem.

Implementation aspects

Fig. 10 is a schematic block diagram of a transmitting node 100 according to the present invention. This node 100 comprises means for communicating with other nodes in the wireless network, represented in the figure by a transmitter/receiver or transceiver 110. This transceiver 110 is particularly suitable for transmitting and receiving data packets and other messages, such as forwarding instructions and inquiry messages, respectively, to external nodes. In addition, the transceiver 110 receives quality data in DQI form and optionally other forms of quality data, such as CQI and routing cost, from external nodes.

A data processing unit 120 is comprised in the transmitting node 100 for processing transmitted and received data. This processing unit 120 typically includes encoder/decoder and modulator/demodulator functionality. In addition, this unit 120 may provide different control messages, such as the query message and forwarding instructions described herein.

The transmitting node 100 further comprises at least one transmit queue or buffer 140 containing data to be transmitted or forwarded to a receiving or relay node in the network. This transmit queue 140 preferably also comprises DQI estimates that the transmitting node has received from other nodes. If the DQI value is associated with, i.e. estimated from, a data packet that the transmitting node has previously transmitted and is thus present in the queue 140, the DQI value is preferably stored in association with the packet. The term "associatively stored" means in this description that the DQI values are stored in such a way that they can be retrieved later on the basis of knowledge of the associated data packet. A typical example of associative storage is when the DQI value and the data packet are stored together as one data entry in the queue 140. Furthermore, the DQI value and the data packet may be stored in different locations in the queue 140 or in two different memories, as long as there is a connection, such as a pointer, between the different memory locations.

The transmitting node 100 preferably also stores the DQI value it has received, e.g. due to snooping, and estimated from data packets not transmitted by the node 100 itself. Node 100 may use this DQI data if it is then responsible for forwarding the packet further.

As briefly explained above, the DQI data may alternatively be stored in another storage location, such as a dedicated quality parameter memory, which may store CQI values, qoS data and route cost metrics in addition to DQI data.

Fig. 11 is a schematic block diagram illustrating one possible implementation of transmit queue 140. Two data streams  _d1 And  _d2 Represented in this example queue 140. First stream  _d1 Involving delivery to a target node v _d1 The data of (1). In this flow, two data packets DP ₁ 、DP ₂ Has been delivered to the stream  _d1 First division ofGroup DP ₁ Multiple relay nodes v with unsuccessful decoding _h 、v _h+1 . Thus, these nodes v _h 、v _h+1 The packet DP which is associated with the first packet DP1 and represents them has been reported ₁ Corresponding decodable quality information DQI _1，h 、DQI _1，h+1 . Quality information DQI _1，h 、DQI _1，h+1 Preferably associated with the data DP in the queue shown in fig. 11 ₁ Are listed together in the table. When e.g. by selecting a candidate receiving node v _h+1 And/or make up this stream  _d1 Second packet DP of ₂ (select packet DP ₂ The type of data contained in) to schedule this stream  _d1 Has generally used this received quality information DQI by the transmitting node _1，h 、DQI _1，h+1 。

And this second packet DP ₂ Associated DQI data DQI _2，h+1 Has been from a node v _h+1 Is received and is present in queue 140. Third packet DP ₃ Recently, this data has been structured but has not yet been transmitted to the relay node or has not yet received its DQI.

Correspondingly, the second stream  _d2 Including to be sent to a target node v _d2 The data of (1). This stream  _d2 First packet DP of ₁ Has been sent to the target node v _d2 And DQI _1，d2 Has already been selected from that node v _d2 Receives and inputs to transmit queue 140. The DQI data can then be provided by the transmitting node, e.g., by selecting between the identity bits and the redundancy bits, to provide the stream  _d2 Is used for the second packet of (1).

As can be seen in fig. 11 and derived from the above discussion, a portion of the data in the transmit queue 140 will have DQIs associated with it, while some data will not have any DQIs at all (since they were not sent or DQIs not reported). In addition to DQI, the record can keep track of how many times the corresponding data has been sent and which HARQ format (e.g. CC, partial IR (PIR) or Full IR (FIR)) has been used.

Returning to fig. 10, the transmitting node 100 further comprises a decision processor or selecting means 130 which performs quality based (DQI based) scheduling and/or routing of data according to the invention. In a first embodiment, this decision processor 130 is configured to select at least one receiving node among a plurality of candidate nodes based on DQI. The relevant DQI used by processor 130 may have been received at one or more different reporting occasions. In other embodiments, the processor 130 selects, based on the DQI, the type of data the data packet will contain and/or which data flow and data packet represented in the transmit queue 140 should be picked up and forwarded first. Combinations of these embodiments, possibly together with DQI-based selection of transmission and/or link parameters, also fall within the scope of the present invention.

In addition to DQI in the scheduling and selection process, the decision processor 130 can use other quality data, such as CQI metrics, routing cost metrics and/or QoS data.

In a preferred embodiment of the invention, the decision processor 130 is configured to perform DQI-based scheduling by optimizing an objective function comprising DQI and optionally CQI, routing cost, qoS data as input parameters.

The

units

110, 120 and 130 of the transmitting node 100 may be provided as software, hardware or a combination thereof. The transmitting node 100 may in turn be arranged in a wireless communication network or system comprising a cellular system as well as a multihop system.

Fig. 12 is a schematic block diagram of another embodiment of a transmitting node 100 according to the present invention. The transmitting node 100 of fig. 12 mainly comprises a conventional receiver chain 110A connected to an antenna or antenna system, a conventional transmit chain 110B (with associated antenna/antenna system), a demodulation and decoder unit 120A, an encoder and modulation unit 120B, a unit 130 for performing DQI-based decision procedures to select data type, relay node, data stream and optional link mode and transmit parameters, a transmit buffer 140, an encapsulation unit 150, a transmit parameter controller 160, an interrogation/probing unit 170, a unit 180 for providing multi-hop cost information, and a receive buffer 190.

In the first round (1), data packets are taken out of the transmit queue 140 and provided to the encapsulation unit 130 for encapsulation and (explicit and/or implicit) addressing. From an addressing perspective, the transmitting node 100 may transmit the data packet using unicast, multicast, or broadcast. The encapsulated data packet is passed to the encoder and modulation unit 120B for encoding 124B and modulation 122B, and also to the transmit chain 110B for transmission to the relay candidate node. The transmit power level and/or antenna weights for the transmission are provided by the transmit parameter controller 160.

In a second round (2), the transmitting node 100 receives response messages from multiple relay candidate nodes via the receiver chain 110A and the unit 120A for demodulation 122A and decoding 124A. Response messages as ACK and/or DQI are then typically passed to decision unit 130 and/or to transmit queue 140. Upon receiving the ACK/DQI, decision processor 130 will perform DQI-based scheduling in accordance with the invention.

However, if other quality parameters are to be used in the scheduling and selection process, the DQI may first be stored in the transmit queue 140 until other such parameters have been provided to the decision processor 130.

In the present example, the decision processor 130 will use cost metrics and CQI data in addition to DQI. Therefore, the query message should be provided and transmitted to the candidate node.

Thus, in the third round (3), the interrogation probe is passed from unit 170 to encapsulation unit 130 for encapsulation and (explicit and/or implicit) addressing. The transmitting node 100 typically transmits the query message to the selected relay candidate node in the network using broadcast or multicast. The relay candidate nodes may be selected, for example, by a general controller (not shown) based on the multi-hop cost information obtained from the underlying route determination protocol 180, possibly along with other information, such as DQI. The encapsulation challenge probe is passed to the encoder and modulation unit 120B and also to the transmission chain 110B for transmission to the relay candidate node. The transmit power level and/or antenna weights for the transmission are provided by the transmit parameter controller 160. In this case, the decision processor 130 may have informed the controller 160 about the appropriate transmission parameters that have been generated from the received DQI data.

In a fourth round (4), the transmitting node 100 receives response messages from multiple relay candidate nodes via the receiver chain 110A and the unit 120A for demodulating 122A and decoding 124A. The response message is passed to decision unit 130. In addition, the relevant DQI data is retrieved from the transmit queue 140 or from some other (temporary) storage means and input to the decision processor 130. Information about the destinations/flows represented in the node 100 and multi-hop cost information from the underlying route determination protocol 180, such as bellman-ford or similar, are also preferably input to the processor 130. In the transmitting node 100, such cost information is preferably collected and/or generated in a multi-hop cost information unit 180 connected to the decision unit 130. Information about the optional destinations and/or flows may be retrieved, for example, by checking the transmit queue 140 or by maintaining a separate list of destinations/flows currently existing in the node 100.

The decision processor 130 then selects at least one of the data type, the relay node, the data flow (and the data packet), and optionally the link mode and transmission parameters for transmission in the decision process. Preferably, the decision unit 130 performs a joint selection of at least two of the above objects. This (joint) selection may be performed by optimizing an objective function based on the DQI data and optionally cost progress and link performance information (CQI), as detailed earlier.

In the fifth round (5), the selected data (type) is then passed from the transmit queue 140 to the encapsulation unit 150, and the encapsulation unit 150 encapsulates the data and sets the address to the selected relay node. The encapsulated packet information is then passed to the encoder and modulation unit 120B, which performs encoding and modulation according to the selected link mode before the packet information is transmitted to the selected relay node using the selected transmission parameters.

The units 110-130 and 150-180 of the transmitting node 100 may be provided as software, hardware or a combination thereof.

Fig. 13 is a schematic block diagram of a communication (relay) node 200 according to the present invention. This node 200 comprises means for communicating with other nodes in the wireless network, represented in the figure by a transmitter/receiver or transceiver 210. This transceiver 210 is particularly suited for transmitting and receiving data packets and other messages to and from external nodes. In addition, the transceiver 210 transmits quality data in the form of DQI and optionally other forms of quality data, such as CQI and routing cost, to external nodes.

A data processing unit 220 is comprised in the communication node 200 for processing transmitted and received data. This processing unit 220 typically includes an encoder and decoder 224A and modulator/demodulator functionality. In addition, this unit 220 may provide different control messages, such as ACK and forward ACK messages as described herein.

The communication node 200 further comprises a DQI generator 230, preferably connected to or at least in communication with the decoder unit 224A of the data processor 220. This generator 230 estimates the degree of decodability of the data packet that was unsuccessfully decoded by the decoder 224A. This degree of decodability may be represented by a number between 0 and 1, representing an estimate of the percentage of data packets or blocks that may be processed (decoded). Alternatively, the degree of decodability may be an estimate of the remaining redundancy required for successful decoding of an unsuccessfully decoded data packet or some other possible DQI measure as described above. Rather than providing a DQI measure, this generator 230 may be configured to produce quality information from which the DQI measure can be derived. In either case, the DQI metric or quality information is provided to the transceiver 210 for reporting (unicast, multicast or broadcast) to the transmitting node.

DQI generator 230 is preferably configured to also evaluate DQI values of data packets intercepted by transceiver 210, i.e. originally directed to other communication nodes.

The units 210 to 230 of the communication node 200 may be provided as software, hardware or a combination thereof. The communication node 200 may in turn be disposed in a wireless communication network or system including a cellular system and a multihop system.

Fig. 14 is a schematic block diagram of relevant parts on the receiver side (receiving communication node 200) according to an exemplary embodiment of the present invention. The relay node 200 of fig. 14 mainly comprises a conventional receiver chain 210A connected to an antenna or antenna system, a conventional transmitter chain 210B (with associated antenna/antenna system), a unit 220A for demodulation 222A and decoding 224A, a unit 220B, DQI generator 230 for modulation 222B and encoding 224B, a transmit queue 240, an encapsulation unit 250, a link performance estimator 260, a transmit node identifier unit 260, and a receive buffer 290.

The relay candidate node 200 receives data packets from one or more transmitting nodes in the wireless network through the receiver chain 210A in round (1). The data packet is provided to a demodulator 222A and a decoder 224A, which attempt to demodulate and decode the packet. If successful, the DQI/ACK generator 230 or a separate acknowledgement unit composes an ACK message through the encapsulation 250 and coding/modulation 220B unit and transmitter chain 210B and returns to the relevant transmitting node.

However, if the packet cannot be successfully decoded by decoder 224A, information indicative of the degree of decodability or the unsuccessfully decoded packet itself is provided to DQI generator 230 for producing a DQI estimate. In addition, unsuccessful packets or data related thereto, such as APP values or other soft information, are stored in the receive buffer 290.

The DQI estimates from DQI generator 230 are provided to encapsulation unit 250 for encapsulation and addressing (the associated address can be obtained from the transmitting node identifier 270). The encapsulated DQI information is then transmitted (unicast, multicast or broadcast) to the relevant transmitting node by using the unit 220B for encoding 224B and modulation 222B and the transmission chain 210B.

In this illustrative embodiment, the transmitting node replies by transmitting an inquiry message captured by the receiver chain 210. In round (2A), the link performance estimator 260 estimates the link performance CQI, such as SNR/SINR (or converts the SNR/SINR value to a supported rate), for transmission back to the inquiring node in a response message. The estimates are passed to the encapsulation unit 250 for encapsulation and addressing. The encapsulated response information is then transmitted to the inquiring transmitting node by using the unit 220B for encoding 224B and modulation 222B and the transmit chain 210B.

If the transmitter address is contained in the inquiry message, the message is also passed to the receive buffer 290 in round (2B) via unit 220A for demodulation 222A and decoding 224A. The transmitting node identifier unit 270 checks the received inquiry message and extracts the transmitter address for passing to the encapsulating unit 250. The transmitter address may then be used by the encapsulation unit 250 to cause the response message to reach the inquiring transmitting node.

If the relay candidate node 200 is chosen by the inquiring transmitting node based on the reported DQI and preferably based on the reported CQI, the relay node 200 typically receives a second data packet from the transmitting node over the receiver chain 210A. This second packet is processed in a similar manner as for the first packet in round (1). In this case, however, decoder 224A may use this received data along with corresponding previously received data present in receive buffer 290 in order to increase the probability of successful decoding. If the data still cannot be decoded, a second DQI, which is now based on the degree of decodability of the combined data of the first and second packets, is generated by the DQI generator 230 and returned to the transmitting node. Accordingly, correctly decoding the data triggers the provision and reporting of ACKs.

The data packets in the receive buffer 290 may then be passed to the transmit queue 240 for further transmission to relay candidate nodes in the multi-hop network. Alternatively, the fully decoded packet is provided to an application (not shown) in the node (200).

The units 210-230 and 250-270 of the communication node 200 may be provided as software, hardware or a combination thereof.

It will be understood by those skilled in the art that various modifications and changes may be made to the present invention without departure from the scope thereof, which is defined by the appended claims.

Reference to the literature

[1] Roongta and j.m.shea, "reliability-based hybrid ARQ with convolutional code", proc.2003 IEEE int.conf.on commun., vol.4, pages 2889-2893, month 5 2003

[2] Roongta, j. -q. Moon and j.m. Schea, "reliability-based hybrid ARQ for partial time interference channels", IEEE mlcom 2004, accepted for presentation, pages 1-7

[3] International application WO 00/21236

[4] International application WO 02/35779

[5] International application WO 2004/091155

[6] International application WO 2004/091154

[7] D.divalar, S.dolinar and F.Pollar, "iterative turbo decoder analysis based on density evolution", IEEE journal on Selected Areas in Communications, vol.19, no.5, pp.891-907, month 5 2001

[8] U.S. Pat. No.6449490

[9] Bender, p.black, m.grob, r.padovani, n.sindhushayana and a.viterbi, "CDMA/HDR: bandwidth efficient high speed wireless data service for streaming users ", IEEE Communications major, vol.38, no.7, pages 70-77, month 7 of 2000.

Claims

1. A method of forwarding data packets in a wireless network (1), the method comprising the steps of, in a transmitting node (100):

-receiving quality information (30) associated with at least one unsuccessfully decoded data packet (10) from at least one communication node (200) of a set of multiple candidate receiving nodes (200), the quality information (30) representing a degree of decodability of the at least one unsuccessfully decoded data packet (10) at the at least one communication node (200);

-selecting at least one receiving node from the set of multiple candidate receiving nodes (200) in dependence of the quality information (30); and

-forwarding a data packet (20) to the receiving node.

2. The method according to claim 1, characterized by receiving said quality information (30) from said at least one receiving node.

3. The method according to claim 1 or 2, wherein said receiving step comprises the step of receiving quality information (30) associated with said at least one unsuccessfully decoded data packet (10) from a plurality of communication nodes (200) of said set of a plurality of candidate receiving nodes (200), said quality information (30) representing a degree of decodability of said at least one unsuccessfully decoded data packet (10) on said plurality of communication nodes (200).

4. The method according to any of the claims 1 to 3, wherein the forwarded data packet (20) comprises data related to data contained in at least one of the at least one unsuccessfully decoded data packet (10).

5. The method according to any of the claims 1 to 4, wherein the selecting step comprises jointly selecting i) the at least one receiving node and ii) at least one of:

-the data packet (20) among a plurality of data packets represented in a transmit queue (140) of the transmitting node (100);

-a type of data in the data packet (20); and

-at least one transmission or link parameter, and said step of forwarding said data packet (20) is performed in accordance with said selected at least one transmission or link parameter.

6. A method of forwarding data packets in a wireless network (1), the method comprising the steps of, in a transmitting node (100):

-receiving quality information (30) associated with at least one unsuccessfully decoded data packet (10) from at least one communication node (200), the quality information (30) representing a degree of decodability of the at least one unsuccessfully decoded data packet (10) at the at least one communication node (200);

-selecting a type of data in a data packet (20) depending on the quality information (30), the data being related to data contained in at least one of the at least one unsuccessfully decoded data packet (10); and

-forwarding said data packet (20) to at least one of said at least one communication node (200).

7. The method of claim 5 or 6, wherein the type of data is selected to be at least one of:

-redundant data; and

-at least a part of the data contained in at least one of the at least one unsuccessfully decoded data packet (10).

8. The method according to claim 6 or 7, wherein the selecting step comprises jointly selecting, based on the quality information (30): i) The type of said data, and ii) at least one transmission or link parameter, and said step of forwarding said data packet (20) is performed in accordance with said selected at least one transmission or link parameter.

9. A method of forwarding data packets in a wireless network (1), the method comprising the steps of, in a transmitting node:

-receiving quality information (30) associated with at least one unsuccessfully decoded data packet (10) from at least one communication node (200), said quality information (30) representing degree of decodability of said at least one unsuccessfully decoded data packet (10) at said at least one communication node (200);

-selecting a flow among a plurality of flows represented in the transmitting node (100) depending on the quality information (30);

-providing data packets (20) from a transmission queue (140) of the transmitting node (100) according to the selected flow, the data packets (20) comprising data related to data contained in at least one of the at least one unsuccessfully decoded data packet (10); and

10. The method of claim 9, wherein the selecting step comprises jointly selecting i) the streams and ii) at least one of:

-a type of data in the data packet (20); and

11. The method according to any of the claims 1 to 10, wherein the selecting step is performed as a function of i) the quality information (30) and ii) at least one of:

-information (50) representative of a link performance between the transmitting node (100) and at least one of the at least one communication node (200);

-a routing cost associated with at least one of the transmitting node and at least one of the at least one communication node (200); and

-at least one quality of service parameter.

12. The method according to any of the claims 1 to 11, wherein the at least one unsuccessfully decoded data packet (10) has been forwarded by the transmitting node (100) previously.

13. The method according to any of the claims 1 to 12, wherein the quality information (30) represents a residual redundancy required for a successful decoding of the at least one unsuccessfully decoded data packet (10).

14. The method according to any of the claims 1 to 13, wherein the selecting step is performed according to an optimization of an objective function comprising the quality information (30).

15. A transmitting node (100) comprising:

-a receiver (110, 110a) for receiving quality information (30) associated with at least one unsuccessfully decoded data packet (10) from at least one communication node (200) of a set of multiple candidate receiving nodes (200), the quality information (30) representing a degree of decodability of the at least one unsuccessfully decoded data packet (10) on the at least one communication node (200);

-means (130) for selecting at least one receiving node from the set of a plurality of candidate receiving nodes (200) in dependence of the quality information (30); and

-a transmitter (110, 110b) for forwarding data packets (20) to the receiving node.

16. The node according to claim 15, wherein the quality information (30) is received from the at least one receiving node.

17. The node according to claim 15 or 16, wherein the receiver (110, 110a) is configured for receiving quality information (30) associated with the at least one unsuccessfully decoded data packet (10) from a plurality of communication nodes (200) in the set of a plurality of candidate receiving nodes (200), the quality information (30) representing a degree of decodability of the at least one unsuccessfully decoded data packet (10) on the plurality of communication nodes (200).

18. The node according to any of the claims 15 to 17, wherein the forwarded data packet (20) comprises data related to data contained in at least one of the at least one unsuccessfully decoded data packet (10).

19. The node according to any of the claims 15 to 18, wherein said selecting means (130) is configured for jointly selecting i) said at least one receiving node and ii) at least one of the following, depending on said quality information (30):

-transmitting the data packet (20) among a plurality of data packets represented in a queue (140);

-a type of data in the data packet (20); and

-at least one transmission or link parameter, and the transmitter (110, 110b) is operable to forward the data packet (20) according to the selected at least one transmission or link parameter.

20. A transmitting node (100) comprising:

-a receiver (110, 110a) for receiving quality information (30) associated with at least one unsuccessfully decoded data packet (10) from at least one communication node (200), said quality information (30) representing a degree of decodability of said at least one unsuccessfully decoded data packet (10) on said at least one communication node (200);

-means (130) for selecting, in dependence on said quality information (30), a type of data in the data packets (20), said data being related to data contained in at least one of said at least one unsuccessfully decoded data packet (10); and

-a transmitter (110, 110b) for forwarding the data packet (20) to at least one of the at least one communication node (200).

21. The node according to claim 20, wherein the selecting means (130) is configured for jointly selecting, in dependence of the quality information (30): i) The type of the data, and ii) at least one transmission or link parameter, and the transmitter (110; 110B) Is operable to forward the data packet (20) in accordance with the selected at least one transport or link parameter.

22. A transmitting node (100) comprising:

-means (130) for selecting a flow among a plurality of flows represented in the transmitting node (100) in dependence on the quality information (30);

-means (220, 250) for providing data packets (20) from a transmission queue (140) of the transmitting node (100) according to the selected flow, the data packets (20) comprising data related to data contained in at least one of the at least one unsuccessfully decoded data packets (10); and

23. The node according to claim 22, wherein said selecting means (130) is configured for jointly selecting, based on said quality information (30), i) said streams and ii) at least one of:

-at least one quality of service parameter; and

24. The node according to any of the claims 15 to 23, wherein said selecting means (130) is configured for performing said selecting based on i) said quality information (30) and ii) at least one of:

-a routing cost associated with at least one of the transmitting node (100) and at least one of the at least one communication node (200); and

-at least one quality of service parameter.

25. The node according to any of the claims 15 to 24, wherein said selecting means (130) is configured for optimizing an objective function comprising said quality information (30).

26. A communication node (200:

-a receiver (210, 210a) for receiving a data packet (10) transmitted from a transmitting node (100) and intended for a receiving node (200-1);

-means (220;

-means (230) for generating quality information (30) indicative of a degree of decodability of said data packet (10) when said decoding means (220; and

-a transmitter (210, 210a) for forwarding the quality information (30) to the transmitting node (100).

27. The communication node according to claim 26, wherein the receiver (210.

28. The communication node according to claim 26 or 27, wherein the transmitter (210.